Meta has revised its upcoming Commercial Terms to remove language that explicitly permitted the company to use content crawled from business-associated websites to “train” its models. The modification follows warnings from industry groups and legal experts regarding the potential for the social media giant to scrape news content without compensation.
The updated terms, scheduled to take effect on October 30, 2026, initially included a clause that granted Meta broad rights to “access, index and crawl” domains associated with business accounts for purposes including “to train.” While the word “train” has been deleted, the revised language still allows Meta to use scraped data for “advertising and business products,” a shift from the broader “Meta Company Products” category found in the original draft. Meta maintains that the update does not change existing crawling practices.
Legal analysts have highlighted that despite the removal of the word “train,” the terms retain a provision allowing Meta to create “test credentials” for automated accounts. These credentials can be used to access, index, and crawl associated domains. James Grimmelmann, a legal expert who scrutinized the original language, warned that such provisions could theoretically allow Meta to bypass paywalls or registration requirements under the guise of “integrity” testing or product maintenance. This mechanism remains a point of concern for digital publishers who rely on subscription models to fund their operations.
The News/Media Alliance criticized the initial policy as an “abuse of market power,” noting that the terms presented businesses with a forced ultimatum. Under the current structure, site owners cannot opt out of these crawling permissions if they wish to maintain a business presence on Facebook, Instagram, or WhatsApp. This creates a friction point for organizations that use Meta’s platforms for distribution but do not want their proprietary content ingested for the company’s internal developments.
Meta maintains that its automated crawlers respect standard robots.txt files, the industry-standard method for website owners to signal which parts of their site should not be indexed. The company asserts that the new Commercial Terms do not override these existing technical protocols. However, the use of contract language to define crawling rights complicates the relationship between a site’s technical settings and its legal obligations as a Meta business user.
The revision comes as Meta continues to expand its AI capabilities, including the development of its Meta Muse model. While Meta has established licensing agreements with some major outlets like CNN and Fox News, the broader news industry remains wary of “hidden” contract language that could potentially allow for the mass harvesting of data without specific per-publisher agreements.
